Applications
在 X 的时间线上，给每条帖子标出它想让你干什么。判断来自 Jev，一个只返回概率、不生成文本的模型。
const OFFICIAL_URL = "https://api.typesafe.ai/v1/systemone";The link points at the commit we read, so the line number still holds.
These question sets are lifted from this project's source exactly as written.
from extension/background.js:12
intentchoiceWhat is `post.text` mainly trying to do?
inform — 传达信息或消息，读者能从中得到点什么。persuade — 论证一个立场，试图改变读者的看法。provoke — 目的是激起愤怒或对立反应。sell — 推广商品、服务或作者自己的东西。entertain — 逗乐、分享日常、社交性质。other — 以上都不比一般情况更贴切。rage_baittrue/falseIs `post.text` engineered to provoke an angry reaction rather than to inform?
synthetictrue/falseDoes `post.text` read as machine-generated at volume rather than written by a person?
undisclosed_adtrue/falseDoes `post.text` promote something commercially without disclosing that it is an ad?
import { experimental_evaluate as evaluate } from 'ai';
const { answers } = await evaluate({
model: 'typesafe-ai/jev',
state,
questions: {
intent: { type: 'choice', instructions: 'What is `post.text` mainly trying to do?',
criteria: { inform: '传达信息或消息，读者能从中得到点什么。', persuade: '论证一个立场，试图改变读者的看法。', provoke: '目的是激起愤怒或对立反应。', sell: '推广商品、服务或作者自己的东西。', entertain: '逗乐、分享日常、社交性质。', other: '以上都不比一般情况更贴切。' } },
rage_bait: { type: 'boolean', instructions: 'Is `post.text` engineered to provoke an angry reaction rather than to inform?' },
synthetic: { type: 'boolean', instructions: 'Does `post.text` read as machine-generated at volume rather than written by a person?' },
undisclosed_ad: { type: 'boolean', instructions: 'Does `post.text` promote something commercially without disclosing that it is an ad?' },
},
});